corona health – a study- and sensor-based mobile app
TRANSCRIPT
Article
Corona Health – A Study- and Sensor-based Mobile App PlatformExploring Aspects of the COVID-19 Pandemic
Felix Beierle 1,∗, , Johannes Schobel 2, Carsten Vogel 1, , Johannes Allgaier 1, Lena Mulansky 1, Fabian Haug 3, JulianHaug 1, Winfried Schlee 4, Marc Holfelder 5, Michael Stach 3, Marc Schickler 3, Harald Baumeister 6, CarolineCohrdes 7, Jürgen Deckert 8, Lorenz Deserno 9, Johanna-Sophie Edler 7, Felizitas A. Eichner 1, Helmut Greger 10, GritHein 11, Peter Heuschmann 1, Dennis John 12, Hans A. Kestler 13, , Dagmar Krefting 14, Berthold Langguth 4, PatrickMeybohm 15, Thomas Probst 16, Manfred Reichert 3, Marcel Romanos 9, Stefan Störk 17, Yannik Terhorst 6, MartinWeiß 11, and Rüdiger Pryss 1
Citation: Beierle, F.; Schobel, J.; Vogel,
C.; Allgaier, J.; Mulansky, L.; Haug, F.;
Haug, J.; Schlee, W.; Holfelder, M.;
Stach, M.; Schickler, M.; Baumeister, H.;
Cohrdes, C.; Deckert, J.; Deserno, L.;
Edler, J.-S.; Eichner, F.A.; Greger, H.;
Hein, G.; Heuschmann, P.; John, D.;
Kestler, H.A.; Krefting, D.; Langguth,
B.; Meybohm, P.; Probst, T.; Reichert,
M.; Romanos, M.; Störk, S.; Terhorst, Y.;
Weiß, M.; Pryss, R. Corona Health – A
Study- and Sensor-based Mobile App
Platform Exploring Aspects of the
COVID-19 Pandemic. 2021.
1 Institute of Clinical Epidemiology and Biometry, University of Würzburg, Germany;[email protected], [email protected], [email protected],[email protected], [email protected], [email protected], [email protected],[email protected]
2 DigiHealth Institute, Neu-Ulm University of Applied Sciences, Germany; [email protected] Institute of Databases and Information Systems, Ulm University, Germany; [email protected],
[email protected], [email protected], [email protected] Department of Psychiatry and Psychotherapy, University Regensburg, Germany; [email protected],
[email protected] LA2 GmbH, Erlangen, Germany; [email protected] Department of Clinical Psychology and Psychotherapy, Ulm University, Germany;
[email protected], [email protected] Mental Health Research Unit, Department of Epidemiology and Health Monitoring, Robert Koch Institute Berlin,
Germany; [email protected], [email protected] Center of Mental Health, Department of Psychiatry, Psychosomatics and Psychotherapy, University Hospital
Würzburg, Germany; [email protected] Department of Child and Adolescent Psychiatry, University Hospital Würzburg, Germany; [email protected],
[email protected] Service Center Medical Informatics, University Hospital Würzburg, Germany; [email protected] Translational Social Neuroscience, University Hospital Würzburg, Germany; [email protected],
[email protected] Lutheran University of Applied Sciences Nürnberg, Germany; [email protected] Institute of Medical Systems Biology, Ulm University, Germany; [email protected] University Medical Center Göttingen, Germany; [email protected] Department of Anaesthesiology, Intensive Care, Emergency and Pain Medicine, University Hospital Würzburg,
Germany; [email protected] Danube University Krems, Austria; [email protected] Comprehensive Heart Failure Center, University and University Hospital Würzburg, Germany [email protected]* Correspondence: [email protected]
Abstract: Physical and mental well-being during the COVID-19 pandemic is typically assessed viasurveys, which might make it difficult to conduct longitudinal studies and might lead to data sufferingfrom recall bias. Ecological momentary assessment (EMA) driven smartphone apps can help alleviatesuch issues, allowing for in situ recordings. Implementing such an app is not trivial, necessitates strictregulatory and legal requirements, and requires short development cycles to appropriately react toabrupt changes in the pandemic. Based on an existing app framework, we developed Corona Health, anapp that serves as a platform for deploying questionnaire-based studies in combination with recordingsof mobile sensors. In this paper, we present the technical details of Corona Health and provide firstinsights into the collected data. Through collaborative efforts from experts from public health, medicine,psychology, and computer science, we released Corona Health publicly on Google Play and the AppleApp Store (in July, 2020) in 8 languages and attracted 7,290 installations so far. Currently, five studiesrelated to physical and mental well-being are deployed and 17,241 questionnaires have been filled out.Corona Health proves to be a viable tool for conducting research related to the COVID-19 pandemicand can serve as a blueprint for future EMA-based studies. The data we collected will substantiallyimprove our knowledge on mental and physical health states, traits and trajectories as well as its riskand protective factors over the course of the COVID-19 pandemic and its diverse prevention measures.
arX
iv:2
106.
0338
6v2
[cs
.CY
] 6
Jul
202
1
2 of 19
Keywords: mobile health; ecological momentary assessment; digital phenotyping; longitudinal studies;mobile crowdsensing
1. Introduction
During the COVID-19 pandemic, it is not only the SARS-CoV-2 coronavirus that affectspublic health, but also the measures taken in trying to contain the spread of the virus, likelockdowns, home office, and further social distancing measures. Existing studies about themental health of the population during the COVID-19 pandemic are typically done via surveys[1–11]. Utilizing smartphone apps for conducting such research has several advantages overweb-based surveys: (1) It is more feasible to conduct longitudinal studies and track users’health conditions over time. (2) Through in situ recordings, recall bias can be avoided. (3)Apps allow for mobile sensing to additionally record objective measurements about the user’ssurroundings and phone usage. (4) Apps offer additional ways of interacting with users, forexample providing feedback that might increase participants’ study engagement. However,developing an app for tracking the mental and physical health of the population is challenging.In the case of tracking public health during the COVID-19 pandemic, the three main challengesare: (a) Such an app has to be developed quickly; it has to be robust and easily adaptable. (b)All legal and regulatory aspects have to be complied with. (c) A methodologically sound andvalid assessment strategy needs to be implemented, able to provide new insights on mentaland physical health during the COVID-19 pandemic.
The pandemic emerged suddenly and time for developing a sophisticated mobile ap-plication was very limited. The typical challenges of mobile software development includefrequent updates to the underlying operating systems, a variety of different development al-ternatives, and a large diversity of devices, sensors, and features. With psychologists, medicaland public health experts, software engineers, computer scientists, and privacy experts, wehave many stakeholder from a variety of fields whose requirements have to be met. Whiletechnical frameworks specifically for mHealth (mobile Health) scenarios exist [12–16], thisdoes not mean that they are off-the-shelf solutions for complex scenarios and requirementsfrom diverse stakeholders. A recent survey on mHealth frameworks listed features regardingmulti-language support, privacy, and customization of data collection as mostly open issues[16]. In addition to that, with the increasing ubiquity of the smartphone and its use for medicalpurposes, the laws and regulations to be compliant with are increasing [17,18]. The EuropeanUnion recently introduced the EU Medical Device Regulation (MDR), that allows registeringan app as a medical product, which was done for our app Corona Health.1
The goal of Corona Health is to track mental and physical well-being during the COVID-19 pandemic. The app allows working interdisciplinary with experts from different fieldsto deploy different questionnaires, and also includes mobile sensing of sensor data andsmartphone usage statistics. Corona Health was initiated by the Mental Health Research Unitof the Robert Koch Institute, the German federal agency for public health responsible fordisease control and prevention. Technically, Corona Health is built with the TrackYourHealthplatform [19], that is successfully utilized in long-running studies about tinnitus, stress, anddiabetes [20–22]. The development time of Corona Health was only 3 months, includingregulatory and legal aspects as well as approval by the ethics committee of the Universityof Würzburg. Its finalization within 3 months shows that the TrackYourHealth platformincluding its established procedures are proper instruments to develop study- and sensor-based mHealth apps. The results presented here may help other researchers working in similarfields, and can serve as technical reference for future study results based on Corona Health.
The main contributions of this paper are:
1 https://www.corona-health.net
3 of 19
• Introduction of the Corona Health app platform and its technical details.• Presentation of statistics about data collected with Corona Health from July 27, 2020 to
May 11, 2021.
In this paper, we present the app’s architecture and its features, illustrate the database,the data exchange between app and backend, detail the study procedure (user journey), andgive insights into the mobile sensing features. Special attention is paid to how Corona Healthfosters the exchange between computer scientists and healthcare professionals. For this,we present the content pipeline, which is used for multi-language support, feedback, anddeployment of new studies. Additionally, we give a brief introduction into dealing with themedical device regulation. Corona Health has been released on Google Play and the AppleApp Store on July 27, 2020. At the time of writing, we count 7,290 installations and 17,241filled-out questionnaires overall. We present the app as well as statistics about the collecteddata.
The paper is structured as follows. In Section 2, relevant background information andrelated works are discussed. Section 3 introduces the technical details of Corona Health,starting with the user’s perspective. Section 4 introduces the data collected so far and providesdescriptive statistics, which are then discussed in Section 5.
2. Background and Related Work
Ecological Momentary Assessment (EMA) is a research method that focuses on assessingmomentary behavior, feelings, or symptoms, typically via self-reports [23,24]. EMA is oftenapplied in psychological research. Using mobile devices for EMA allows for higher validity[25] and filling out a questionnaire in situ helps avoid recall bias [26]. Additionally, mobiledevices, typically smartphones, are equipped with several different sensors and can yieldstatistics about when and how the users interacts with the device. Collecting such contextdata [27] is called mobile sensing. Generally, we can distinguish between two types of sensing.Participatory sensing focuses on the active collection and sharing of data by users [28]. Here,the user actively decides about the sensing process. Opportunistic sensing on the other handrefers to passive sensing with minimal user involvement [29]. In the context of this paper,filling out a questionnaire can be regarded as participatory sensing, and getting sensor data likethe user’s location can be considered opportunistic sensing.
Mobile crowdsensing (MCS) refers to the measurement of large-scale phenomena thatcannot be measured by a single device [30–32]. This can be, for example, measuring thenoise-level across a city, or, like in the context of this paper, measuring the mental andphysical well-being of people during the ongoing COVID-19 pandemic. MCS in the healthcaredomain can bring its own challenges, like specific regulations and providing feedback tousers [31,33]. The combination of EMA and MCS allows for so-called digital phenotyping[34–39]. Digital phenotyping describes the idea of extending the concept of a phenotype tothe behavior of people in relation to digital devices and services, in particular smartphones.Digital phenotyping research is conducted, for example, in personality science [40,41], formental health assessment [42], or for predicting mood changes [43].
There are several technical frameworks for the development of MCS apps that supportEMA, for example AWARE [12] or Sensus [13]. In a recent survey paper, Kumar et al. reviewedsuch frameworks, specifically for mHealth studies and applications [16]. Kumar et al. definethree types of stakeholders – researchers, developers, and end users. Only two frameworkssupport all three of the defined stakeholders. There are further differences between thesurveyed frameworks. There is a large range of offered features and functionalities, forexample with respect to the types of sensor data that can be collected. Perhaps, due to theeffort associated with it, most of the reviewed frameworks are not maintained after release.
As open challenges, the authors of [16] list, among others, a lack of internationalizationsupport. Only one of the reviewed frameworks supports surveys in multiple languages. Wesuspect that such kind of shortcomings lead to the development of new or specialized frame-
4 of 19
works. As another open challenge, Kumar et al. describe the personalization of data collectionand the handling of user consent, which with most frameworks have to be implemented bythe developers utilizing the framework.
Although users in general are quite willing to share data with researchers [44], especiallyfor COVID-19 related apps, privacy is a major concern [45,46]. Typically, the use of apps forstudies has to be approved by institutional review boards, and developers have to comply withlocal privacy-related laws like the GDPR (General Data Protection Regulation) in the EuropeanUnion. On top of that, additional local or regional laws and regulations can bring additionalrequirements for researchers and developers. Germany introduced regulations about medicaldevices, which includes strict regulations for apps relating to healthcare. Effectively, theseregulations allow apps to be approved as medical products. Additionally, both Google Playand the Apple App Store introduced special requirements for COVID-related apps.
We built Corona Health based on our own framework platform TrackYourHealth, tacklingthe shortcoming of existing frameworks. We support multiple languages and comply to alllocal laws, including the German medical product regulations for healthcare-related apps.
3. Technical Details
In this section, we present the technical details of Corona Health. Corona Health wasbuilt with the TrackYourHealth platform and API [19,47] and consists of a PHP-based backendand two native apps (Android and iOS). The management of content is performed througha content pipeline developed by the team over years. In future, a web-based application isscheduled to manage the pipeline more conveniently. In the following, we start by describingthe contents of Corona Health from a user perspective (Section 3.1). In Section 3.2, we detailthe overall architecture and features from a technical perspective. Section 3.3 describes thedatabase schema and what data is recorded. Section 3.4 summarizes how the communicationbetween mobile apps and backend works. Section 3.5 describes the mobile sensing that isconducted in Corona Health. Section 3.6 details how data gets into the Corona Health systemfrom a backend perspective, for example, for integrating new studies. This integration ofnew studies allows for further collaboration with new research partners. We conclude thetechnical description of Corona Health with a brief overview of how compliance with themedical device regulations in Germany was implemented (Section 3.7). While Corona Health’stechnical base can serve as a blueprint for similar EMA apps, its concrete implementation andrelease process cover some aspects specifically related to COVID-19. This includes releaserequirements like the strict guidelines set by the app store providers, and this includes featureslike the curated coronavirus-related news tab.
3.1. User Perspective
Users can download the Corona Health app from Google Play and the Apple App Storefor free. The app can be found via the search functionality of the stores, and the app wasadvertised on Twitter and in newsletters. Figure 1 shows the flow of using Corona Healthfrom a user perspective. The first time the app is used, the app shows the disclaimer (see 1©in Fig. 1). Without consent, the app will be closed. After consenting, the user is asked aboutif he/she wants to allow notifications ( 2© in Fig. 1; iOS shown here). The Studies Tab thenshows the five currently available studies. Each study that the user selects to participate incontains an initial baseline questionnaire. The user selects each of those in order to open it.At this point, the app asks the user if he/she wants to allow the mobile sensing features tobe activated. Point 3© in Fig. 1 shows this for iOS (location) and for Android, additionallyto location, app usage statistics can be chosen to be shared. We present more details aboutmobile sensing in Section 3.5. Filling out the baseline questionnaire concludes the onboardingphase of Corona Health.
After onboarding ( 4© in Fig. 1), the user has access to the five main features of the app:Questionnaires, Info, News, Studies, and About Us. The follow-up questionnaire are scheduled
5 of 19
App Usage after Onboarding
Afte
r On
bo
ard
ing
Questionnaires
IFeedback to
Questionnaires
IIWeekly News
IIIStudy Selection
IVFurther
Information
V
First time at tab Studies with first questionnaire
See Open
Questionnaires
Select
Questionnaire
Allow Location
Measurements?
Follow-Up Questionnaires(with schedule)
3
4
First-time Procedure1
Click StartUser
Consent?
App will be closed
Direct User to Studies
Route to Study Tab2
Studies Tab
Allow Notificatons?
Studies Tab
Mental health for
adults
Physical health
for adults
Mental health for
adolescents
Recognizing
stress for adults
University
Medicine Network:
Compass Project
On
bo
ard
ing
Corona Health User Journey
Figure 1. User Journey of Corona Health, highlighting the steps for initial start of the app, registration for studies, filling out baseline andfollow-up questionnaires.
6 of 19
by the system. The Info tab contains feedback based on the filled-out questionnaires. The Newstab contains weekly news curated by selected members of the provided studies. Through theStudies tab, the user can sign up for additional studies that he/she did not sign up for yet.The About Us tab contains settings, Corona Health team information, and all legal policies.Figure 2 shows screenshots of both the Android and the iOS version of Corona Health. Figure3 shows the flow for filling out questionnaires, the core feature of Corona Health, focusing onnotifications. After the baseline questionnaire, the notification schedule for the selected studyis activated. Based on that schedule, notifications will be triggered. Additionally, the user canalways fill out additional follow-up questionnaires by manually accessing them in the app.
3.2. App Architecture and Modules
Figure 4 is a flowchart depicting the processes in the app from a technical perspective.The continuous lines show user interactions and the dashed lines show data interactions in thebackground. After installation of Corona Health, the app can be started and an anonymouslogin process is started. We developed the app with offline functionality in mind. If thebackend server cannot be reached, the local database temporarily handles the login, storageof questionnaires and mobile sensing data.
Figure 5 shows the core features of Corona Health, divided into backend and mobiledevice modules. The backend modules are handled by the admin. The user can interact withthe modules on the mobile device. The management of the notifications is done by the admin.
3.3. Database Schema
Corona Health uses a relational database to store and manage both user-generated andoperational data. Since data integrity and consistency mechanisms (e.g., constraints and ACIDtransactions) are already integrated with relational databases, highly inter-related data modelsare easier to create and maintain with this type of database system. Furthermore, the querylanguage SQL is suitable for complex analytical queries needed for the data evaluation. Inorder to give insights into the project’s database schema, Fig. 6 depicts an Entity-RelationshipModel (ERM) using Martin’s notation. Note that Fig. 6 illustrates an excerpt of the entiredatabase schema. The selection represents the core entities containing most of the user-generated data. Due to the multilingual project setup, the database schema contains severalentities with translated text data, marked with the red T©, which are omitted in the ERM tofurther simplify the model.
To describe the data set in this work, a snapshot was extracted on May 11, 2021. CoronaHealth combines MCS and EMA to collect user data. For this reason, the entity Usersconstitutes a central and high-related table with 7,290 verified users. In addition, the users’actions (e.g., study subscription) are stored in the table User Histories to give insights intothe app usage. Currently, there are 4,495 subscriptions of 2,802 unique users to 5 multilingualstudies. To manage the systems’ studies, so-called collaborators are users with additional(role-based) permissions on specific studies.
To collect data in a structured way, one study has one or more questionnaires in theentity of the same name. The latter contains 10 unique questionnaires in each language of thecorresponding study. Questionnaires are structured with polymorphic buildings blocks calledQuestionnaire Elements (n = 1, 276). On average, each questionnaire is structured with 255elements that can be of the type Page Element (n = 117), Text Element (n = 159), QuestionElement (n = 976), Headline Element (n = 24) or Media Element (n = 0). The submittedanswers for a questionnaire is serialized and stored in JSON in the entity Answersheets (n= 17,241). The latter also store sensor data (e.g., location) and client device information (e.g.,operating system) in the same table. In order to give the user feedback immediately aftersubmission of a questionnaire, such a questionnaire may reference to one or more key-rulepairs that are stored in the entity Feedback (n = 54). Rules, evaluated on the client side, canbe managed and adjusted dynamically. Feedback translations can also be stored here.
7 of 19
Android Version
iOS Version
Figure 2. Screenshots of the iOS and Android versions of Corona Health.
8 of 19
Users
Fill out Baseline
Questionnaires
Fill out Follow-Up
QuestionnairesSubscribe
Fill out Follow-Up
Questionnaires
8
Fill out Follow-Up
Questionnaires
15
Fill out Follow-Up
Questionnaires
1
Fill out Follow-Up
Questionnaires
10
Users can always trigger the
questionnaire themselves in
addition to schedule
Based on
initially
configured
schedule
Activate scheduleEvery Friday
4pm Example
schedule
Figure 3. Notification scheduling in Corona Health. After the baseline questionnaire is filled out, the notification schedule for the selectedstudy is activated. Notifications will be triggered based on the given schedule. Additionally, the user can always fill out additionalfollow-up questionnaires by manually accessing them.
Questionnaires StudiesNewsInfo
(Feedback) About
Study Selection
Login
Start App
Installation of App
Allow Location Measurement?Allow UsageEvents (only Android)?
Server
localDB
online
offlineUser DataAnonym. ID
offline
online
offline
online
Figure 4. Corona Health processes of the User Journey on a technical level. Continuous lines indicate user interactions and dashed linesindicate background data interactions. After installation, the app can be started and an anonymous login process is started. If the backendserver cannot be reached, the local database temporarily handles the login, storage of questionnaires and mobile sensing data.
9 of 19
Export Data Visualize Results Manage News
Manage Sensors Provide Feedback
User Management Manage Studies
Manage Baseline
Questionnaires
Manage Follow-Up
Questionnaires
Apply Notifications Sensor Data
Subscribe to Studies Provide Feedback
Fill in Baseline
Questionnaires
Fill in Follow-Up
Questionnaires
Show Additional Information
Manage Notifications
Backend
Mobile Device
Figure 5. The core modules of Corona Health.
UsersUserHistories Studies
Roles
Collaborators
Answersheets Questionnaires
Feedback
QuestionnaireElements
MediaElement
PageElement
QuestionElement
HeadlineElement
TextElement
T
T
T
T
T
T
T
T
Figure 6. Excerpt of Corona Health’s relational database schema. The circled T© indicates the presence of translated text data.
10 of 19
3.4. Data Exchange
In the course of developing Corona Health, a RESTful backend service was developed.This backend, in turn, serves as a common API for all clients (i.e., smart mobile devicesand web applications). The server application was developed using the well-known Laravelframework and relies on JSON:API2 as a specification for building APIs for exchanging data.This specification, in turn, increases productivity by structurally describing request andresponse formats. JSON:API describes how a client application should request and modifyresources, and how a server should response to such requests.
The backend server offers routes to access resources from the data model (see Fig.7). In particular, resources denoted in the Design Time part can be accessed by clients. Forexample, details about Studies, or the Questionnaires assigned to Studies can be requested.A Questionnaire, in turn, comprises different Elements, like Headlines, Texts, or actualQuestions to collect data respectively.
Study Questionnaire Feedback
User Resultset Evaluation
1
1
n
n
1
1
n
1
n
n
1
n
1
n
Element
1
n
Headline Pagebreak
Question
MediaText
Design Time
Run Time
Figure 7. API Data Model.
Finally, healthcare professionals may define Feedback rules that may be used to automati-cally evaluate filled-out Questionnaires. Clients, in turn, can submit data via RESTful endpointsfor resources described in the Run Time part. In particular, when answering Questionnaires,a Resultset is created. For each Resultset, an Evaluation based on the previously definedFeedback can be requested. In addition to this automated feedback, the API also offers a directpatient-to-doctor communication via a dedicated messaging API (which is not shown in Fig.7).
By its architecture design and the use of JSON:API the backend service can be easilyadapted to future requirements. For example, a psycho-education module was recentlyimplemented and further modules are planned.
3.5. Mobile Sensing
Corona Health collects mobile sensing data that is relevant in the context of conducingresearch related to the physical and mental well-being of the population during the COVID-19pandemic. The data we collect are device information (Android and iOS), coarse locationdata during questionnaire fill-out (Android and iOS), and aggregated app usage statistics(Android only). All of these data have been shown to be relevant for research. For example,for EMA apps, it has been shown that the used OS should be used as a covariate in analyses[48]. Location data has been shown to have some predictability for depressive mood [49] and
2 https://jsonapi.org/; last accessed: 2021-03-15
11 of 19
positive affect [50]. The averages for app usage statistics have been shown to be different forusers with different scores on the Big Five personality trait scales [51].
Each time a questionnaire is filled out, device info, and OS version are stored. If the usergives his/her permission, we also query for and store the current location. The permission forthe location is handled by the operating system (Android or iOS). We do not store the rawlocation, but process the data that we get from the OS to make it more coarse grained, andonly keep a value with an accuracy of 11.1 km. The level of detail with which we store themobile sensing data was determined through discussions with all stakeholder, institutionalreview board, and the relevant laws and regulations.
On Android, if the user gives the relevant permission, it is possible to access the user’sapp usage history via UsageEvents3. The data allows a very detailed look into the users’interactions with his/her smartphone. Before storing data locally and sending it to thebackend, we aggregate some statistics and only store: (1) overall daily phone usage; (2) timeperiods of activity and inactivity; (3) for some apps: daily sum of usage; timestamps of firstand last usage. The apps for which we store some usage statistics are the combination of afixed list of social media apps and the five most used apps in terms of screen time. Tables 1and 2 give a detailed overview of the data being collected.
Table 1: Data collected about app usage in Corona Health (Android).
Field CommentbeginTime Beginning of observed period
endTime End of observed period
collectedAt Timestamp of observation time
apps List of apps and their usage statistics (see Table 2)
top5Apps Top 5 most used apps and their usage statistics (see Table 2)
sleepTimes List of tuples with beginning and end timestamps of time windows in which the devicewas sleeping for at least one hour (no active screen)
screenTime Contains two lists with daily cummulated time values. List 1: use time of all apps(screen on with app in visible foreground). List 2: time in which the screen was active(including no app being used or woken up from a notification).
3.6. Content Pipeline
This section describes the process of bringing the content of questionnaires into theapp. In each of the four stages, experts with different domain knowledge are included. Thisincludes app developers, medical professionals, psychologists, etc. Existing publications alsoshow that such tight communication procedures are highly needed [52]. The entire process forCorona Health, in turn, is shown in Fig. 8. In the first stage, the questionnaire is designed in adiscussion group. At the same time, information is provided about the technical frameworkso that it can be incorporated into the questionnaire design. This involves, for example,information on how certain types of questions can be technically represented (e.g., slider, etc.).
In order to create a questionnaire, an Excel template is used. It has two sheets labeledBaseline and FollowUp. This corresponds to the principle of one sheet per questionnaire. Ineach of the sheets, one line describes one element in the questionnaire. This element, in turn,could be either a headline, a textfield, a pagebreak, or a question. For each question, the creatorof the questionnaire has to define which type it has, if it is optional, what the variable nameis, and how the answer options should be encoded. Different encodings are highlighted indifferent colors. The encodings and number of response options must be consistent across all
3 https://developer.android.com/reference/android/app/usage/UsageEvents, last accessed: 2021-03-18
12 of 19
Excel Template•Defines metadata for questionnaire
•Supports multiple languages
•Determines question types and variable type
•Created by Non-IT experts with domain knowledge
Converter script•Converts the excel template into an API-readable format (.xlsx to .json)
•Extracts information from the excel file row by row
•Tests the structure of the excel file for common mistakes
API•Imports in different JSON files
•Offers RESTful APIs for mobile clients to access questionnaires and meta data
App•Requests data from API•Questionnaires are only shown in the language of choice
Feedback Loop
Test
Figure 8. Steps to publish a questionnaire within the mobile application. The automated conversion allows for fast iteration cycles, resultingin a more flexible and end-user driven development process.
Table 2: Values stored for each app that is being recorded (Android).
Field CommentpackageName Name of the package
completeUseTime Sum of daily use time
completeFGServiceUseTime Sum of ForegroundService use time (app is running without beingvisible on the screen)
dailyValues List of containers for the following values and timestamps. OneContainer per day.
useTime Time the app was in the foreground (visible on the screen)
firstUseTime Timestamp of first visible use
lastUseTime Timestamp of the last visible use
FGServiceUseTime Sum of ForegroundService use time (app is running without beingvisible)
firstFGServiceUseTime Timestamp of first ForeGroundService use
lastFGServiceUseTime Timestamp of (start of) last ForeGroundService use
languages. If a certain question is repeated in the FollowUp questionnaire, the template getsall the content of this question using a unique variable name and cross referencing.
The converter script (second step in Fig. 8; i.e., the second stage) splits up the Excelsheet into two types of subsets. The first type contains the meta data about the questionnaire,the second type contains the questionnaire itself with one subset per language. Depending onthe monitor scaling or version of Excel user, unwanted spaces or line breaks are generated,which then lead to incorrect or unintended display of the questions when called up in the app.For the language subsets, the converter script parses each line and cleans up the strings
13 of 19
by removing whitespaces or unintentional line breaks. It finally saves the questionnaire asa .json file. In the third stage, a seeder script stores the JSON-file into the backend, therebyreflecting already existing versions (through versioning). After the seeder script has beenexecuted, in the fourth stage, the API is ready to operate and communicate with the mobileapplications.
3.7. Medical Device Regulation
In general, more and more countries require to develop mHealth applications compliantto the medical device regulation (MDR). When adhering to the MDR, the important aspectconstitutes the validation of the software, which shall ensure that a software is workingcorrectly over time, i.e., always generating the same output for the same input. Therefore,the entire system has been validated on the basis of the IEC 62304 and IEC 82304 standards(medical device software/healthcare apps) as well as the GAMP 5 regulations (standard workof the pharmaceutical industry) [53]. These standards are the relevant parts of the medicaldevice regulation (MDR) for implementing software that is compliant with the MDR. Thebasis for such development is a harmonized, risk-based approach developed from theseregulations. Essentially, the regulations requires the preparation of several document types,which finally serve as the basis for the validation of the software [53]. To be more precise,in a first step, the requirements of the application were collected in detail, in order to havea foundation to implement these requirements. The implementation took place in an agileprocess: design – programming – testing. As soon as all requirements have been implementedand tested in various development cycles, they were finally tested in a system test. The actualapp development speed was decelerated by the relatively long requirement phase. However,this detailed analysis of the requirements and regulatory specifications led to an efficient errorand problem identification and evaluation process. Therefore, during the further developmentcourse, potential error sources could be identified and eliminated rapidly.
4. Collected Data
At first, we list the descriptions of the provided studies to illustrate the informationintroduced to participants:
• Study on Mental health for adults (18 years +): People’s health is very important to us.Hence, we would like to ask you something: How are you feeling these days? Whilephysical illness is prominent in news and the public, we are interested in the mood andpossible stress factors that influence it. Moving freely, daily work routine, eating out,visiting grandparents, or meeting friends are important ingredients for us to feel good.Consequently, we would like to learn more about how you maintain social contacts, asthese are important factors related to your well-being. In order to limit the effort for youas much as possible, we would like to automatically record your social contact via yoursmartphone (e.g. number and duration of calls, text messages). We would also like toautomatically record your location at the time of the survey (GPS signal) in order to takeregional differences into account. The data as well as your answers are stored completelyanonymously, to profile you as a person is not possible. This first survey is a little moreextensive and takes about 20 minutes - the follow-up surveys are only half of the length.By taking the time and participating in this survey you are a great help!
• Study on Mental health for adolescents (12 to 17 years): This scientific study investigates theburden on adolescents worldwide aged 12-17 years during the coronavirus-pandemic.Understanding your struggles may help us physicians and scientists to develop strategiesto better cope with the pandemic. Would you like to take part and help us? Thefirst survey takes 15 minutes. You may then take part in weekly follow-up surveys (5minutes). The study is completely anonymous. The study has been approved by anethics committee and by a data protection official. No tracking is done by the app and
14 of 19
only the data that you agree to will be transmitted anonymously. If you are unsure,whether you should take part, please talk to your parents about it. THANK YOU!
• Study on Physical health for adults (18 years +): Many things in our daily life have changedsince the Corona crisis. This also includes more basic things such as our diet or our freetime for example, how we do sports. These are factors that also have an influence on thedevelopment and course of diseases such as cardiovascular diseases. In the followingquestionnaire, we would thus like to learn more about how the Corona crisis affects yourhabits with a focus on factors that are related to the cardiovascular system. Filling inthe questionnaire only takes about 15 minutes. All data is collected anonymously. Itis therefore not possible to relate any of the collected data with an individual person.Thank you for your support!
• Study on Recognizing stress for adults (18 years and up): Recognizing stress is importantfor your mental and physical health. That is why we invite you to fill out a shortstress questionnaire once a week. People react very differently to stress. Therefore,we are also interested in factors that influence the experience of stress such as yourage, your family status, your gender, and your smartphone usage (e.g. frequency ofusing communication services and social networks). The smartphone usage is recordedautomatically to minimize the effort for you. Likewise, your whereabouts at the time ofthe survey are recorded automatically (GPS signal) in order to take regional differences inthe stress experience into account. The data as well as your answers are stored completelyanonymously. No conclusions can be drawn about the content of your social interactionsor your person. Thank you very much for your effort and time in answering the questionsabout your perceived stress. You are a great help!
• Study on University Medicine Network: Compass project on the acceptance of pandemic apps(18 years and older).: The variety of pandemic apps, apps that have been developed, forexample, in hackathons to control COVID-19, shows the great potential that many expertssee in them. But for apps to be effective in the pandemic, they must be used by manypeople. This applies not only to the Corona warning app, but also in particular to appsfor assessing individual risk, for example in the case of certain pre-existing conditions.It is therefore necessary that such apps enjoy trust among the general population andthat data can be analyzed together for medical research with the consent of the users.In COMPASS, scientists from a wide range of disciplines from university hospitals arejoining forces with partners from science and industry in an interdisciplinary project tojointly develop a coordination and technology platform for pandemic apps. Help uswith this survey so that pandemic apps can be developed even better and with broadacceptance and transparency in the future.
As of May 11, 2021, a total of 2,802 users have registered in the five studies. Theycompleted a total of 17,241 questionnaires. The study participants are between 12 and 120years old (mean: 40.34, SD: 14.75).4 The ratio of Android to iOS users is about 4 to 1 acrossall studies. From 7,290 verified users, only 2,802 filled-out any questionnaire (38.4%). Highdrop-out rates are a well-known issue in EMA studies [54]. From those 2,802 users, 1,476 filledout follow-up questionnaires (52.7%). On average, each of those 1,476 filled out 8.6 follow-upquestionnaires. We provide more statistics about the collected data in Table 3.
5. Discussion
In this paper, we presented the Corona Health app. It was conceived and implementedto enable observational studies that are able to explore aspects of the COVID-19 pandemicusing smartphones. Based on the presented figures, we concluded that the app is frequentlyused and valuable data could be gathered. However, the data we collected likely contains a
4 The outlier with an age of 120 years likely did not enter the correct age.
15 of 19
Table 3: Fact Sheet for the five studies. A total of 17,241 questionnaires were completed by May 11, 2021. Note thatdemographic information was not available for the Adult Physical Health Questionnaire (Study "Phys. > 18 yrs") until theDecember 2020 update.
Study name Compass Stress Phys. > 18 yrs Psych. > 18 yrs Psych. < 18 yrs
n baselines 314 371 1479 2141 190
n followups not applicable 1514 3588 7289 355
Females % 59.9% 65.2% 61.5% 50.5% 40.1%
Male % 37.6% 33.4% 38.3% 48.9% 56.5%
Diverse % 1.3% 1.3% 2.8% 0.7% 1.7%
Not answered 1.3% 0.0% 0.0% 0.0% 1.7%
Age Distribution *
(12, 13] 0 0 0 0 21
(13, 14] 0 0 0 0 29
(14, 15] 0 0 0 0 33
(15, 16] 0 0 0 0 46
(16, 17] 1 0 0 0 48
(17, 30] 79 99 97 531 0
(30, 40] 71 100 82 566 0
(40, 50] 65 79 64 453 0
(50, 60] 49 62 69 380 0
(60, 70] 31 29 38 171 0
(70, 80] 5 2 8 37 0
(80, 120] 1 0 0 4 0
Android 258 303 315 1365 124
iOS 44 68 43 777 31
Location perm. 80.25% 77.90% 81.74% 79.93% 61.58%
App perm. (Android) 49.17% 43.00% 32.49% 29.19% 29.94%
selection bias and the sample might not be representative of the general population [55]. Abroader picture of society would have been desirable, but is challenging to achieve. CoronaHealth has been developed within 3 months and it is compliant with the Medical DeviceRegulation. To enable this, we have shown that Corona Health is based on existing technologyas well as well established procedures to manage the collaboration between the study domainexperts and computer scientists. Practically, the following features constitute the key aspectsin this context. They are summarized to show our experiences in developing apps like CoronaHealth in a short period of time with many regulatory and organizational constraints:
• Our API can be flexibly used to manage various study types.• Our apps can be flexibly tailored to the needs of researchers from the respective healthcare
domain. Experience gained from previous projects, how such apps shall operate whenapplied large scale in real-life proved very valuable for the design procedure.
• Our proven and accepted content pipeline is able to flexibly add and change contentbetween healthcare experts and computer scientists (see Section 3.6).
• We were able to find regulatory experts that are experienced in the MDR in the contextof software engineering projects.
• Our general user journey of the app is accepted by participants.
Furthermore, for the feature to gather data about the used apps of Android participants,it can be concluded that it is a valuable addition. Participants mainly allow the measurementof this type of data and first results indicate that in the context of well-being and mental healthapp usage statistics might reveal interesting results. For example, a publication based onCorona Health data has found the following: Younger participants with higher use times
16 of 19
tended to report less social well-being and higher loneliness, while the opposite effect wasfound for older adults [56].
To conclude, we constitute the framework that enables us to implement solutions likeCorona Health with its discussed features as a flexible and robust instrument. However, inpractice, we also learned that many aspects are not covered so far. First, our questionnairemodule is not able to dynamically navigate participants through a questionnaire. For example,if questions are not relevant for a specific participant, all dependent questions still have tobe answered or inspected. Second, we identified scenarios, in which our feedback modulewould require a more complex rule engine to cover all feedback scenarios. Third, a powerfulvisualization feature to enable participants viewing their own data directly within the appwould be highly welcome. Finally, a feature should be offered that provides an extendedonboarding procedure when the app is used for the first time (e.g., by the presentation of avideo that explain the app). Altogether, Corona Health is now running for almost one year andhas revealed its usefulness for exploring aspects of the COVID-19 pandemic. Currently, weenhance our framework by new modules and improve the convenience with respect to contentmanagement. After that, other sensor measurements and more advanced interventions areplanned (e.g., just-in-time interventions).
Author Contributions: Conceptualization, J.S., C.V., J.A., L.M., F.H., J.H., M.H., H.B., C.C., J.D. J.-S.E.,F.A.E., H.G. G.H., P.H., D.J., T.P., M.Ro., W.S., Y.T. and R.P.; data curation, J.A., M.St. and R.P.; formalanalysis, J.A. and R.P.; funding acquisition, R.P; investigation, J.S., C.V., L.M., M.H., H.B., C.C., J.-S.E.,F.A.E., P.H., D.J., H.A.K., D.K., B.L., T.P., M.Ro., W.S., S.S., Y.T. and R.P.; methodology, J.S., C.V., J.A., L.M.,F.H., J.H., H.B., C.C., J.D., J.-S.E., F.A.E., G.H., P.H., D.J., D.K., B.L., P.M., T.P., M.Ro., W.S., S.S., Y.T. andR.P.; project administration, J.S., C.V., L.M., M.H., H.B., C.C., J.-S.E., F.A.E., P.H., D.J., B.L., T.P., M.Re.,M.Ro., W.S., S.S., Y.T. and R.P.; resources, H.G., P.H., H.A.K., M.Re., M.Ro., W.S. and R.P.; software, J.S.,C.V., L.M., F.H., J.H., H.G. and R.P.; supervision, J.S., H.B., C.C., J.-S.E., P.H., D.J., M.Re., Y.T. and R.P.;validation, J.S., L.M., M.H., W.S. and R.P.; visualization, J.A., M.St. and M.Sc.; writing—original draftpreparation, F.B., J.S., J.A., L.M., M.St. and R.P.; writing—review and editing, F.B., J.S., C.V., J.A., L.M.,F.H., J.H., M.H., M.St., M.Sc., H.B., C.C., J.D., L.D., J.-S.E., F.A.E., H.G., G.H., P.H., D.J., H.A.K., D.K., B.L.,P.M., T.P., M.Re., M.Ro., W.S., S.S., Y.T., M.W. and R.P. All authors have read and agreed to the publishedversion of the manuscript.
Funding: F.B., L.M., F.H., J.H., M.St., C.C., P.H., D.K., B.L., W.S., and R.P. are supported by grants inthe project COMPASS. P.H. and R.P. are supported by a grant in the project NAPKON. The COMPASSand NAPKON projects are part of the German COVID-19 Research Network of University Medicine("Netzwerk Universitätsmedizin”), funded by the German Federal Ministry of Education and Research(funding reference 01KX2021). L.D. is supported by a grant from the German Research Foundation (TRR265, Project A02, 428318753) and by the Federal Ministry for Education and Research, Germany (FKZ:01EO1501). G.H. is supported by a grant from the German Research Foundation (HE 4566/5-1).
Institutional Review Board Statement: The Corona Health app study was conducted in accordancewith the German medical products law. The data protection officer and the ethics committee of theUniversity of Würzburg, Germany approved the study (No. 130/20-me). The procedures used in thisstudy adhere to the tenets of the Declaration of Helsinki.
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.
Acknowledgments: We are grateful for the script programming provided by Leoni Holl and KlausKammerer, and for the support with the visual design of the app provided by Michael Schultz.
Conflicts of Interest: J.D. is an investigator in the EU-Horizon-funded Predict Study of P1Vital andCo-Applicant with BioVariance in the InDepth Study funded by the Bavarian State Government.
References1. Zhang, Y.; Ma, Z.F. Impact of the COVID-19 Pandemic on Mental Health and Quality of Life among Local Residents in Liaon-
ing Province, China: A Cross-Sectional Study. International Journal of Environmental Research and Public Health 2020, 17, 2381.doi:10.3390/ijerph17072381.
17 of 19
2. Wang, C.; Pan, R.; Wan, X.; Tan, Y.; Xu, L.; Ho, C.S.; Ho, R.C. Immediate Psychological Responses and Associated Factors during theInitial Stage of the 2019 Coronavirus Disease (COVID-19) Epidemic among the General Population in China. International Journal ofEnvironmental Research and Public Health 2020, 17, 1729. doi:10.3390/ijerph17051729.
3. Gualano, M.R.; Lo Moro, G.; Voglino, G.; Bert, F.; Siliquini, R. Effects of Covid-19 Lockdown on Mental Health and Sleep Disturbancesin Italy. International Journal of Environmental Research and Public Health 2020, 17, 4779. doi:10.3390/ijerph17134779.
4. Xiong, J.; Lipsitz, O.; Nasri, F.; Lui, L.M.W.; Gill, H.; Phan, L.; Chen-Li, D.; Iacobucci, M.; Ho, R.; Majeed, A.; McIntyre, R.S. Impact ofCOVID-19 Pandemic on Mental Health in the General Population: A Systematic Review. Journal of Affective Disorders 2020, 277, 55–64.doi:10.1016/j.jad.2020.08.001.
5. Gloster, A.T.; Lamnisos, D.; Lubenko, J.; Presti, G.; Squatrito, V.; Constantinou, M.; Nicolaou, C.; Papacostas, S.; Aydın, G.; Chong,Y.Y.; Chien, W.T.; Cheng, H.Y.; Ruiz, F.J.; Garcia-Martin, M.B.; Obando-Posada, D.P.; Segura-Vargas, M.A.; Vasiliou, V.S.; McHugh, L.;Höfer, S.; Baban, A.; Neto, D.D.; da Silva, A.N.; Monestès, J.L.; Alvarez-Galvez, J.; Paez-Blarrina, M.; Montesinos, F.; Valdivia-Salas,S.; Ori, D.; Kleszcz, B.; Lappalainen, R.; Ivanovic, I.; Gosar, D.; Dionne, F.; Merwin, R.M.; Kassianos, A.P.; Karekla, M. Impact ofCOVID-19 Pandemic on Mental Health: An International Study. PLOS ONE 2020, 15, e0244809. doi:10.1371/journal.pone.0244809.
6. Limcaoco, R.S.G.; Mateos, E.M.; Fernández, J.M.; Roncero, C. Anxiety, Worry and Perceived Stress in the World Due to the COVID-19Pandemic, March 2020. Preliminary Results. medRxiv 2020, p. 2020.04.03.20043992. doi:10.1101/2020.04.03.20043992.
7. Pieh, C.; Budimir, S.; Probst, T. The Effect of Age, Gender, Income, Work, and Physical Activity on Mental Health during CoronavirusDisease (COVID-19) Lockdown in Austria. Journal of Psychosomatic Research 2020, 136, 110186. doi:10.1016/j.jpsychores.2020.110186.
8. Pierce, M.; McManus, S.; Hope, H.; Hotopf, M.; Ford, T.; Hatch, S.L.; John, A.; Kontopantelis, E.; Webb, R.T.; Wessely, S.; Abel, K.M.Mental Health Responses to the COVID-19 Pandemic: A Latent Class Trajectory Analysis Using Longitudinal UK Data. The LancetPsychiatry 2021, 0. doi:10.1016/S2215-0366(21)00151-6.
9. Pieh, C.; Budimir, S.; Delgadillo, J.; Barkham, M.; Fontaine, J.R.J.; Probst, T. Mental Health During COVID-19 Lockdown in the UnitedKingdom. Psychosomatic Medicine 2021, 83, 328–337. doi:10.1097/PSY.0000000000000871.
10. Dale, R.; Budimir, S.; Probst, T.; Stippl, P.; Pieh, C. Mental Health during the COVID-19 Lockdown over the Christmas Period inAustria and the Effects of Sociodemographic and Lifestyle Factors. International Journal of Environmental Research and Public Health2021, 18, 3679. doi:10.3390/ijerph18073679.
11. Budimir, S.; Pieh, C.; Dale, R.; Probst, T. Severe Mental Health Symptoms during COVID-19: A Comparison of the United Kingdomand Austria. Healthcare 2021, 9, 191. doi:10.3390/healthcare9020191.
12. Ferreira, D.; Kostakos, V.; Dey, A.K. AWARE: Mobile Context Instrumentation Framework. Frontiers in ICT 2015, 2.doi:10.3389/fict.2015.00006.
13. Xiong, H.; Huang, Y.; Barnes, L.E.; Gerber, M.S. Sensus: A Cross-Platform, General-Purpose System for Mobile Crowdsensing inHuman-Subject Studies. Proc. 2016 ACM International Joint Conference on Pervasive and Ubiquitous Computing (UbiComp 2016).ACM, 2016, UbiComp ’16, pp. 415–426. doi:10.1145/2971648.2971711.
14. Schobel, J.; Probst, T.; Reichert, M.; Schickler, M.; Pryss, R. Enabling sophisticated lifecycle support for mobile healthcare datacollection applications. IEEE Access 2019, 7, 61204–61217.
15. Schobel, J.; Pryss, R.; Schickler, M.; Ruf-Leuschner, M.; Elbert, T.; Reichert, M. End-user programming of mobile services: empoweringdomain experts to implement mobile data collection applications. 2016 IEEE International Conference on Mobile Services (MS). IEEE,2016, pp. 1–8.
16. Kumar, D.; Jeuris, S.; Bardram, J.E.; Dragoni, N. Mobile and Wearable Sensing Frameworks for mHealth Studies and Applications: ASystematic Review. ACM Transactions on Computing for Healthcare 2020, 2, 8:1–8:28. doi:10.1145/3422158.
17. Hamel, M.B.; Cortez, N.G.; Cohen, I.G.; Kesselheim, A.S. FDA regulation of mobile health technologies. The New England journal ofmedicine 2014, 371, 372.
18. Jogova, M.; Shaw, J.; Jamieson, T. The regulatory challenge of mobile health: lessons for Canada. Healthcare Policy 2019, 14, 19.19. Vogel, C.; Pryss, R.; Schobel, J.; Schlee, W.; Beierle, F. Developing Apps for Researching the COVID-19 Pandemic with the
TrackYourHealth Platform. 2021 IEEE/ACM 8th International Conference on Mobile Software Engineering and Systems (MobileSoft).IEEE, 2021, pp. 65–68. doi:10.1109/MobileSoft52590.2021.00015.
20. Pryss, R.; Schlee, W.; Langguth, B.; Reichert, M. Mobile crowdsensing services for tinnitus assessment and patient feedback. 2017IEEE International Conference on AI & Mobile Services (AIMS). IEEE, 2017, pp. 22–29.
21. Pryss, R.; others. Exploring the time trend of stress levels while using the Crowdsensing Mobile health platform, TrackYourStress, andthe influence of perceived stress reactivity: ecological momentary assessment pilot study. JMIR mHealth and uHealth 2019, 7, e13978.
22. Unnikrishnan, V.; others. Predicting the Health Condition of mHealth App Users with Large Differences in the Number of RecordedObservations-Where to Learn from? International Conference on Discovery Science. Springer, 2020, pp. 659–673.
23. Stone, A.A.; Shiffman, S. Ecological Momentary Assessment (EMA) in Behavioral Medicine. Annals of Behavioral Medicine 1994,16, 199–202. doi:10.1093/abm/16.3.199.
24. Larson, R.; Csikszentmihalyi, M. The Experience Sampling Method. In Flow and the Foundations of Positive Psychology: The CollectedWorks of Mihaly Csikszentmihalyi; Reis, H., Ed.; Springer, 2014; Vol. Flow and the Foundations of Positive Psychology, pp. 41–56.doi:10.1007/978-94-017-9088-8_2.
18 of 19
25. van Berkel, N.; Ferreira, D.; Kostakos, V. The Experience Sampling Method on Mobile Devices. ACM Computing Surveys 2017,50, 93:1–93:40. doi:10.1145/3123988.
26. Pryss, R.; Probst, T.; Schlee, W.; Schobel, J.; Langguth, B.; Neff, P.; Spiliopoulou, M.; Reichert, M. Prospective Crowdsensing versusRetrospective Ratings of Tinnitus Variability and Tinnitus–Stress Associations Based on the TrackYourTinnitus Mobile Platform.International Journal of Data Science and Analytics 2019, 8, 327–338. doi:10.1007/s41060-018-0111-4.
27. Beierle, F.; Tran, V.T.; Allemand, M.; Neff, P.; Schlee, W.; Probst, T.; Pryss, R.; Zimmermann, J. Context Data Categories and PrivacyModel for Mobile Data Collection Apps. Procedia Computer Science 2018, 134, 18–25. doi:10.1016/j.procs.2018.07.139.
28. Burke, J.; Estrin, D.; Hansen, M.; Parker, A.; Ramanathan, N.; Reddy, S.; Srivastava, M.B. Participatory Sensing. Workshop onWorld-Sensor-Web (WSW’06): Mobile Device Centric Sensor Networks and Applications. ACM, 2006, pp. 117–134.
29. Lane, N.D.; Miluzzo, E.; Lu, H.; Peebles, D.; Choudhury, T.; Campbell, A.T. A Survey of Mobile Phone Sensing. IEEE CommunicationsMagazine 2010, 48, 140–150. doi:10.1109/MCOM.2010.5560598.
30. Ganti, R.K.; Ye, F.; Lei, H. Mobile Crowdsensing: Current State and Future Challenges. IEEE Communications Magazine 2011, 49, 32–39.doi:10.1109/MCOM.2011.6069707.
31. Kraft, R.; Schlee, W.; Stach, M.; Reichert, M.; Langguth, B.; Baumeister, H.; Probst, T.; Hannemann, R.; Pryss, R. Combining mobilecrowdsensing and ecological momentary assessments in the healthcare domain. Frontiers in neuroscience 2020, 14, 164.
32. Boubiche, D.E.; Imran, M.; Maqsood, A.; Shoaib, M. Mobile Crowd Sensing – Taxonomy, Applications, Challenges, and Solutions.Computers in Human Behavior 2019, 101, 352–370. doi:10.1016/j.chb.2018.10.028.
33. Pryss, R. Mobile crowdsensing in healthcare scenarios: taxonomy, conceptual pillars, smart mobile crowdsensing services. In DigitalPhenotyping and Mobile Sensing; Springer, 2019; pp. 221–234.
34. Baumeister, H.; Montag, C., Eds. Digital Phenotyping and Mobile Sensing: New Developments in Psychoinformatics; Studies in Neuroscience,Psychology and Behavioral Economics, Springer International Publishing, 2019. doi:10.1007/978-3-030-31620-4.
35. Insel, T.R. Digital Phenotyping: Technology for a New Science of Behavior. JAMA 2017, 318, 1215–1216. doi:10.1001/jama.2017.11295.36. Jain, S.H.; Powers, B.W.; Hawkins, J.B.; Brownstein, J.S. The Digital Phenotype. Nature Biotechnology 2015, 33, 462–463.
doi:10.1038/nbt.3223.37. Montag, C.; Sindermann, C.; Baumeister, H. Digital Phenotyping in Psychological and Medical Sciences: A Reflection about Necessary
Prerequisites to Reduce Harm and Increase Benefits. Current Opinion in Psychology 2020, 36, 19–24. doi:10.1016/j.copsyc.2020.03.013.38. Onnela, J.P. Opportunities and Challenges in the Collection and Analysis of Digital Phenotyping Data. Neuropsychopharmacology 2021,
46, 45–54. doi:10.1038/s41386-020-0771-3.39. Beierle, F. Integrating Psychoinformatics with Ubiquitous Social Networking: Advanced Mobile-Sensing Concepts and Applications; Springer,
2021. doi:10.1007/978-3-030-68840-0.40. Montag, C.; Baumeister, H.; Kannen, C.; Sariyska, R.; Meßner, E.M.; Brand, M. Concept, Possibilities and Pilot-Testing of a New
Smartphone Application for the Social and Life Sciences to Study Human Behavior Including Validation Data from PersonalityPsychology. J 2019, 2, 102–115. doi:10.3390/j2020008.
41. Beierle, F.; Matz, S.C.; Allemand, M. Mobile Sensing in Personality Science. In Mobile Sensing in Psychology: Methods and Applications;Mehl, M.R.; Wrzus, C.; Eid, M.; Harari, G.; Priemer, U.E., Eds.; Guilford Press: New York City, NY, USA, 2021 (to appear).
42. Rooksby, J.; Morrison, A.; Murray-Rust, D. Student Perspectives on Digital Phenotyping: The Acceptability of Using SmartphoneData to Assess Mental Health. Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems. ACM, 2019, CHI’19, pp. 425:1–425:14. doi:10.1145/3290605.3300655.
43. Zulueta, J.; Piscitello, A.; Rasic, M.; Easter, R.; Babu, P.; Langenecker, S.A.; McInnis, M.; Ajilore, O.; Nelson, P.C.; Ryan, K.; Leow, A.Predicting Mood Disturbance Severity with Mobile Phone Keystroke Metadata: A BiAffect Digital Phenotyping Study. Journal ofMedical Internet Research 2018, 20, e241. doi:10.2196/jmir.9775.
44. Beierle, F.; Tran, V.T.; Allemand, M.; Neff, P.; Schlee, W.; Probst, T.; Zimmermann, J.; Pryss, R. What Data Are Smartphone UsersWilling to Share with Researchers? Journal of Ambient Intelligence and Humanized Computing 2020, 11, 2277–2289. doi:10.1007/s12652-019-01355-6.
45. Borra, S. COVID-19 Apps: Privacy and Security Concerns. In Intelligent Systems and Methods to Combat Covid-19; Joshi, A.; Dey, N.;Santosh, K.C., Eds.; SpringerBriefs in Applied Sciences and Technology, Springer, 2020; pp. 11–17. doi:10.1007/978-981-15-6572-4_2.
46. Ahmed, N.; Michelin, R.A.; Xue, W.; Ruj, S.; Malaney, R.; Kanhere, S.S.; Seneviratne, A.; Hu, W.; Janicke, H.; Jha, S.K. A Survey ofCOVID-19 Contact Tracing Apps. IEEE Access 2020, 8, 134577–134601. doi:10.1109/ACCESS.2020.3010226.
47. Pryss, R.; Schobel, J.; Reichert, M. Requirements for a Flexible and Generic API Enabling Mobile Crowdsensing mHealth Applications.2018 4th International Workshop on Requirements Engineering for Self-Adaptive, Collaborative, and Cyber Physical Systems(RESACS). IEEE, 2018, pp. 24–31. doi:10.1109/RESACS.2018.00010.
48. Pryss, R.; Schlee, W.; Hoppenstedt, B.; Reichert, M.; Spiliopoulou, M.; Langguth, B.; Breitmayer, M.; Probst, T. Applying MachineLearning to Daily-Life Data From the TrackYourTinnitus Mobile Health Crowdsensing Platform to Predict the Mobile Operat-ing System Used With High Accuracy: Longitudinal Observational Study. Journal of Medical Internet Research 2020, 22, e15547.doi:10.2196/15547.
19 of 19
49. Canzian, L.; Musolesi, M. Trajectories of Depression: Unobtrusive Monitoring of Depressive States by Means of Smartphone MobilityTraces Analysis. Proceedings of the 2015 ACM International Joint Conference on Pervasive and Ubiquitous Computing. ACM, 2015,UbiComp ’15, pp. 1293–1304. doi:10.1145/2750858.2805845.
50. Heller, A.S.; Shi, T.C.; Ezie, C.E.C.; Reneau, T.R.; Baez, L.M.; Gibbons, C.J.; Hartley, C.A. Association between Real-World ExperientialDiversity and Positive Affect Relates to Hippocampal–Striatal Functional Connectivity. Nature Neuroscience 2020, 23, 800–804.doi:10.1038/s41593-020-0636-4.
51. Beierle, F.; Probst, T.; Allemand, M.; Zimmermann, J.; Pryss, R.; Neff, P.; Schlee, W.; Stieger, S.; Budimir, S. Frequency and Duration ofDaily Smartphone Usage in Relation to Personality Traits. Digital Psychology 2020, 1, 20–28. doi:10.24989/dp.v1i1.1821.
52. Gründahl, M.; Deckert, J.; Hein, G. Three questions to consider before applying ecological momentary interventions (EMI) inpsychiatry. Frontiers in psychiatry 2020, 11, 333.
53. Holfelder, M.; Mulansky, L.; Schlee, W.; Baumeister, H.; Schobel, J.; Greger, H.; Hoff, A.; Pryss, R. Medical Device Regulation Effortsfor mHealth Apps – An Experience Report of Corona Check and Corona Health. arXiv:2104.13635 [cs] 2021.
54. Seifert, A.; Hofer, M.; Allemand, M. Mobile Data Collection: Smart, but Not (Yet) Smart Enough. Frontiers in Neuroscience 2018, 12.doi:10.3389/fnins.2018.00971.
55. Hargittai, E. Potential Biases in Big Data: Omitted Voices on Social Media. Social Science Computer Review 2020, 38, 10–24.doi:10.1177/0894439318788322.
56. Wetzel, B.; Pryss, R.; Baumeister, H.; Edler, J.S.; Gonçalves, A.S.O.; Cohrdes, C. “How Come You Don’t Call Me?” SmartphoneCommunication App Usage as an Indicator of Loneliness and Social Well-Being across the Adult Lifespan during the COVID-19Pandemic. International Journal of Environmental Research and Public Health 2021, 18, 6212. doi:10.3390/ijerph18126212.